Benjamin_Loison/waybackpy
1
0
Fork 0
Code Issues 1 Pull Requests Packages Projects Releases Wiki Activity

test adjusted for new changes

Browse Source
This commit is contained in:
Akash
2020-07-17 20:45:09 +05:30
committed by GitHub
parent cf9c83f3b8
commit 7e847c210b
1 changed files with 32 additions and 26 deletions
Download Patch File Download Diff File Expand all files Collapse all files

58
tests/test_1.py
View File

@@ -3,81 +3,86 @@ sys.path.append("..")
import waybackpy import waybackpy
import pytest import pytest
user_agent = "Mozilla/5.0 (Windows NT 6.2; rv:20.0) Gecko/20121202 Firefox/20.0" user_agent = "Mozilla/5.0 (Windows NT 6.2; rv:20.0) Gecko/20121202 Firefox/20.0"
def test_clean_url(): def test_clean_url():
test_url = " https://en.wikipedia.org/wiki/Network security " test_url = " https://en.wikipedia.org/wiki/Network security "
answer = "https://en.wikipedia.org/wiki/Network_security" answer = "https://en.wikipedia.org/wiki/Network_security"
test_result = waybackpy.clean_url(test_url) target = waybackpy.Url(test_url, user_agent)
test_result = target.clean_url()
assert answer == test_result assert answer == test_result
def test_url_check(): def test_url_check():
InvalidUrl = "http://wwwgooglecom/" broken_url = "http://wwwgooglecom/"
with pytest.raises(Exception) as e_info: with pytest.raises(Exception) as e_info:
waybackpy.url_check(InvalidUrl) waybackpy.Url(broken_url, user_agent)
def test_save(): def test_save():
# Test for urls that exist and can be archived. # Test for urls that exist and can be archived.
url1="https://github.com/akamhy/waybackpy" url1="https://github.com/akamhy/waybackpy"
archived_url1 = waybackpy.save(url1, UA=user_agent) target = waybackpy.Url(url1, user_agent)
archived_url1 = target.save()
assert url1 in archived_url1 assert url1 in archived_url1
# Test for urls that are incorrect. # Test for urls that are incorrect.
with pytest.raises(Exception) as e_info: with pytest.raises(Exception) as e_info:
url2 = "ha ha ha ha" url2 = "ha ha ha ha"
waybackpy.save(url2, UA=user_agent) waybackpy.Url(url2, user_agent)
# Test for urls not allowed to archive by robot.txt. # Test for urls not allowed to archive by robot.txt.
with pytest.raises(Exception) as e_info: with pytest.raises(Exception) as e_info:
url3 = "http://www.archive.is/faq.html" url3 = "http://www.archive.is/faq.html"
waybackpy.save(url3, UA=user_agent) target = waybackpy.Url(url3, user_agent)
target.save()
# Non existent urls, test # Non existent urls, test
with pytest.raises(Exception) as e_info: with pytest.raises(Exception) as e_info:
url4 = "https://githfgdhshajagjstgeths537agajaajgsagudadhuss8762346887adsiugujsdgahub.us" url4 = "https://githfgdhshajagjstgeths537agajaajgsagudadhuss8762346887adsiugujsdgahub.us"
archived_url4 = waybackpy.save(url4, UA=user_agent) target = waybackpy.Url(url3, user_agent)
target.save()
def test_near(): def test_near():
url = "google.com" url = "google.com"
archive_near_year = waybackpy.near(url, year=2010, UA=user_agent) target = waybackpy.Url(url, user_agent)
archive_near_year = target.near(year=2010)
assert "2010" in archive_near_year assert "2010" in archive_near_year
archive_near_month_year = waybackpy.near(url, year=2015, month=2, UA=user_agent) archive_near_month_year = target.near( year=2015, month=2)
assert ("201502" in archive_near_month_year) or ("201501" in archive_near_month_year) or ("201503" in archive_near_month_year) assert ("201502" in archive_near_month_year) or ("201501" in archive_near_month_year) or ("201503" in archive_near_month_year)
archive_near_day_month_year = waybackpy.near(url, year=2006, month=11, day=15, UA=user_agent) archive_near_day_month_year = target.near(year=2006, month=11, day=15)
assert ("20061114" in archive_near_day_month_year) or ("20061115" in archive_near_day_month_year) or ("2006116" in archive_near_day_month_year) assert ("20061114" in archive_near_day_month_year) or ("20061115" in archive_near_day_month_year) or ("2006116" in archive_near_day_month_year)
archive_near_hour_day_month_year = waybackpy.near("www.python.org", year=2008, month=5, day=9, hour=15, UA=user_agent) target = waybackpy.Url("www.python.org", user_agent)
archive_near_hour_day_month_year = target.near(year=2008, month=5, day=9, hour=15)
assert ("2008050915" in archive_near_hour_day_month_year) or ("2008050914" in archive_near_hour_day_month_year) or ("2008050913" in archive_near_hour_day_month_year) assert ("2008050915" in archive_near_hour_day_month_year) or ("2008050914" in archive_near_hour_day_month_year) or ("2008050913" in archive_near_hour_day_month_year)
with pytest.raises(Exception) as e_info: with pytest.raises(Exception) as e_info:
NeverArchivedUrl = "https://ee_3n.wrihkeipef4edia.org/rwti5r_ki/Nertr6w_rork_rse7c_urity" NeverArchivedUrl = "https://ee_3n.wrihkeipef4edia.org/rwti5r_ki/Nertr6w_rork_rse7c_urity"
waybackpy.near(NeverArchivedUrl, year=2010, UA=user_agent) target = waybackpy.Url(NeverArchivedUrl, user_agent)
target.near(year=2010)
def test_oldest(): def test_oldest():
url = "github.com/akamhy/waybackpy" url = "github.com/akamhy/waybackpy"
archive_oldest = waybackpy.oldest(url, UA=user_agent) target = waybackpy.Url(url, user_agent)
assert "20200504141153" in archive_oldest assert "20200504141153" in target.oldest()
def test_newest(): def test_newest():
url = "github.com/akamhy/waybackpy" url = "github.com/akamhy/waybackpy"
archive_newest = waybackpy.newest(url, UA=user_agent) target = waybackpy.Url(url, user_agent)
assert url in archive_newest assert url in target.newest()
def test_get(): def test_get():
oldest_google_archive = waybackpy.oldest("google.com", UA=user_agent) target = waybackpy.Url("google.com", user_agent)
oldest_google_page_text = waybackpy.get(oldest_google_archive, UA=user_agent) assert "Welcome to Google" in target.get(target.oldest())
assert "Welcome to Google" in oldest_google_page_text
def test_total_archives(): def test_total_archives():
count1 = waybackpy.total_archives(" https://google.com ", UA=user_agent) target = waybackpy.Url(" https://google.com ", user_agent)
assert count1 > 500000 assert target.total_archives() > 500000
count2 = waybackpy.total_archives("https://gaha.e4i3n.m5iai3kip6ied.cima/gahh2718gs/ahkst63t7gad8", UA=user_agent) target = waybackpy.Url(" https://gaha.e4i3n.m5iai3kip6ied.cima/gahh2718gs/ahkst63t7gad8 ", user_agent)
assert count2 == 0 assert target.total_archives() == 0
if __name__ == "__main__": if __name__ == "__main__":
test_clean_url() test_clean_url()
@@ -96,3 +101,4 @@ if __name__ == "__main__":
print(".") print(".")
test_total_archives() test_total_archives()
print(".") print(".")
print("OK")